Ivan Bestvina
Verified Expert in Engineering
Data Scientist and AI Developer
Zagreb, Croatia
Toptal member since April 2, 2020
With over seven years of data science and over nine years of coding experience, Ivan has developed a wide range of data and AI products for startups and large companies. He has a proven record of managing development and data science teams and enjoys building products from the ground up. He specializes in large language models, predictive modeling, network science, and anomaly detection and has excellent communication skills and attention to detail.
Portfolio
Experience
- SQL - 9 years
- Python - 9 years
- Network Science - 8 years
- NumPy - 8 years
- Pandas - 8 years
- Data Science - 8 years
- Predictive Modeling - 7 years
- Large Language Models (LLMs) - 3 years
Preferred Environment
Jupyter, Git, Visual Studio Code (VS Code), Python
The most amazing...
...product I've developed was Qluent.com, an AI assistant enabling non-technical business experts to analyze data and leverage insights.
Work Experience
Co-founder
Qluent AB
- Developed the core AI stack of the solution, including the question-to-SQL generation with process explanations.
- Participated in sales and fundraising efforts with another co-founder.
- Developed an AI assistant that enables users to explore their database in natural language from Slack.
Senior Data Scientist
Clas Ohlson AB
- Developed an automated promotions system, including forecasting, cross-sales, and cannibalization analysis modules. Maintained and further developed an automated markdown system.
- Participated in developing a centralized data warehouse, connecting various internal and external data sources. Developed an extensive analytical stack on top of the data warehouse.
- Led AI workshops to upskill different teams across the organization.
Data Scientist
Athena Analytics LTD
- Implemented an end-to-end data pipeline for student score prediction and anomaly detection.
- Delivered in-depth analysis of data patterns and model performance, enabling important strategic data-driven decisions.
- Corrected multiple issues with the previous processing and modeling implementation, improving prediction accuracy while reducing the computation time by more than 200x.
Co-founder/Lead Data Scientist
Clarify
- Co-designed a novel automated BI reporting system that detects shifts and anomalies in retail data and delivers them as natural language news feed posts enriched with interactive charts and data exploration tools.
- Led the research into predictive modeling of the demand of highly perishable produce, as well as the application of such models for pattern and anomaly detection.
- Spearheaded sales efforts and customer relations of the company.
Senior Data Scientist
FoodTracks
- Developed the framework and the processing pipelines for demand forecasting for German bakeries, all the way from the client's ERP data sources to the automated production orders.
- Created a novel data science management framework and, based on it, led a team of data scientists through product research and development.
- Prototyped, developed, and maintained tens of tailor-made data products for retail management.
Software Developer
Mireo
- Developed core C++ algorithms for a platform-agnostic car navigation system.
- Implemented WebSocket server and client in C++ from scratch for use in standardized navigation app notifications.
- Adapted the car navigation system for use in proprietary client hardware.
Data Scientist
Ericsson
- Developed machine learning models for smart home sensors data anomaly detection to detect occupant accidents and assist elders living alone.
- Set up data pipelines for IoT data processing in Java.
- Created exploratory analysis reports comparing the outputs of different smart home sensors.
Student Project Assistant
University of Zagreb
- Developed novel election prediction models based on social network topology data.
- Developed telecom operator churn predictors based on communication networks.
- Modeled drug effectiveness based on virus protein networks.
- Developed epidemic model parameter estimators based on snapshots of infected nodes.
Experience
Datasloth
https://github.com/ibestvina/dataslothDAGpy
https://pypi.org/project/dagpy/Education
Master's Degree in Computer Science
University of Zagreb - Zagreb, Croatia
Bachelor's Degree in Computer Science
University of Zagreb - Zagreb, Croatia
Skills
Libraries/APIs
Pandas, NumPy, Matplotlib, NetworkX, Scikit-learn, XGBoost, SQLAlchemy, PyTorch, REST APIs
Tools
Jupyter, ChatGPT, Haystack, PyCharm, Git
Languages
Python, SQL, Java, C++, JavaScript
Frameworks
LlamaIndex, Django
Platforms
Ubuntu, Visual Studio Code (VS Code), Docker, Google Cloud Platform (GCP)
Storage
PostgreSQL
Other
Data Science, Natural Language Processing (NLP), Artificial Intelligence (AI), Machine Learning, Data Cleansing, Generative Pre-trained Transformers (GPT), Chatbots, Random Forests, Data Analytics, OpenAI GPT-3 API, OpenAI GPT-4 API, Language Models, Retrieval-augmented Generation (RAG), OpenAI, Prompt Engineering, Technical Leadership, Generative Pre-trained Transformer 3 (GPT-3), Software Development, Data Visualization, Network Science, Predictive Modeling, Large Language Models (LLMs), LangChain, Deep Learning, Neural Networks, Llama 2, Web Scraping, Recommendation Systems, ECharts, Software Development Management, Sales Forecasting
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring